Revert "make it work when factors"

This reverts commit ed32638.

Author: strengejacke

R/estimate_grouplevel.R

@@ -342,24 +342,21 @@ estimate_grouplevel.stanreg <- function(
 
   # Analyze random effect structure
   randomgroups <- insight::find_random(model)$random
-
-  # formulas for random effects
-  f_random <- insight::find_formula(model)$random
+  # randomslopes <- insight::find_random_slopes(model)
+  # TODO: currently insight::find_random_slopes() doesn't tell us to which random factor does the slope belong to
+  # So we need to find it manually by callin ranef()
   randomslopes <- list()
-  # iterate random effects
-  for (s in f_random) {
-    # extract slope and group
-    p_slope <- gsub("(.*)\\|(.*)", "\\1", insight::safe_deparse(s))
-    p_group <- gsub("(.*)\\|(.*)", "\\2", insight::safe_deparse(s))
-    # clean
-    re <- all.vars(s)
-    # extract slopes
-    slopes <- insight::trim_ws(re[sapply(re, grepl, p_slope, fixed = TRUE)])
-    # if slopes exist, add them to list
-    if (length(slopes) > 0) {
-      # make sure we use related group name as element name
-      gr <- insight::trim_ws(re[sapply(re, grepl, p_group, fixed = TRUE)])
-      randomslopes[[gr]] <- slopes
+  p <- lme4::ranef(model)
+  for (g in names(p)) {
+    s <- names(p[[g]])[names(p[[g]]) != "(Intercept)"]
+
+    # Only pick non-factor random slopes for now
+    # TODO: correctly deal with the case when the random slope is a factor
+    pred <- insight::find_predictors(model, effects = "all", flatten = TRUE)
+    s <- s[s %in% pred]
+
+    if (length(s) > 0) {
+      randomslopes[[g]] <- s
     }
   }
 

tests/testthat/test-estimate_grouplevel.R

@@ -240,21 +240,6 @@ test_that("estimate_grouplevel type='marginal'", {
     gl3[gl3$Parameter == "hp", "Coefficient"]
   )
   expect_gt(r$estimate, 0.99)
-
-  # works when grouping variables are factors
-  data(mtcars)
-  mtcars$carb <- as.factor(mtcars$carb)
-  mtcars$gear <- as.factor(mtcars$gear)
-
-  model <- lme4::lmer(mpg ~ hp + (1 + hp | carb) + (1 | gear), data = mtcars)
-  gl1 <- estimate_grouplevel(model, type = "random")
-  gl3 <- estimate_grouplevel(model, type = "marginal")
-
-  r <- cor.test(
-    gl1[gl1$Parameter == "(Intercept)" & gl1$Group == "carb", "Coefficient"],
-    gl3[gl3$Parameter == "(Intercept)" & gl3$Group == "carb", "Coefficient"]
-  )
-  expect_equal(r$estimate, 0.6626235, tolerance = 1e-4, ignore_attr = TRUE)
 })
 
 test_that("estimate_grouplevel type='marginal' correlations", {